A system dynamics view of predictive and iterative delivery

Purpose of the work:

This work represents predictive (waterfall) and iterative (agile) project management as conceptual stock-and-flow diagrams (SFDs). The objective is not to simulate project performance or estimate outcomes numerically. Instead, the purpose is to use system dynamics as a visual reasoning language to show how each approach structures work, control, feedback, rework, learning, and value delivery. The core argument is that predictive and agile approaches differ not only in sequencing; they differ in feedback architecture. 

Reference Basis

The project management logic is grounded in PMI standards and guidance. The PMBOK Guide and The Standard for Project Management define project management as a value-oriented discipline that must be adapted to different delivery environments (Project Management Institute, 2026). PMI’s Development Approach and Life Cycle Performance Domain emphasizes the relationship among development approach, delivery cadence, and project life cycle, including predictive, adaptive, and hybrid choices (Project Management Institute, n.d.). The agile model also draws on the Agile Practice Guide, developed by PMI in collaboration with the Agile Alliance, which provides guidance on applying agile and hybrid approaches (Project Management Institute, 2017). The modeling logic draws from system dynamics, particularly the use of stocks, flows, feedback loops, and delays to understand dynamic complexity (Sterman, 2001).

Method Used to Create the Models

The modeling method was conceptual and comparative. First, each approach was bounded from authorized work to accepted value or deliverables. Second, major accumulations were identified as stocks: Approved Scope, Designed Work, Built Work, Tested Work, and Accepted Deliverables in the predictive model; and Product Backlog, Sprint Backlog, Work in Progress, Increment, Released Value, Customer Insight, Team Learning, Technical Debt, and Rework Backlog in the agile model. Third, the transformation of work between these stocks was represented as flow rates. Fourth, the major feedback loops were added to clarify how each approach responds to change, quality problems, schedule pressure, stakeholder feedback, and learning. Because this is an explanatory model, the diagrams intentionally prioritize clarity over equation-level completeness. See the following Stock Flow Diagrams (SFD) that represent each approach.

Comparison of the feedback structures

Analysis

The predictive model is best understood as a phase-gated control system. Work moves through a planned sequence, and governance is applied through scope approval, progress monitoring, variance detection, and corrective action. This structure is useful when requirements are stable, compliance matters, and coordination can be planned up front. 

Its weakness is feedback delay: defects, requirement misunderstandings, or integration issues may not become visible until testing or acceptance, when corrections are more disruptive. The key reinforcing risk is the schedule pressure-quality erosion-rework loop, where pressure can reduce quality discipline, increase defect escape, and create additional delay.

The agile model is best understood as a short-cycle learning system. Work moves through backlog refinement, sprint commitment, work in progress, increment creation, and release/acceptance. The central distinction is that feedback is deliberately accelerated: released value generates customer insight, retrospectives build team capability, and WIP limits stabilize flow. 

Agile is not simply “flexible”; the model shows that it depends on discipline, as reflected in the Definition of Ready, Definition of Done, WIP control, prioritization quality, and technical debt management. Its failure mode is not delayed feedback alone, but unmanaged feedback, excessive WIP, or debt accumulation that lowers velocity and increases rework.

Conclusion and Closing Notes

The comparison suggests that predictive and agile approaches should be viewed less as competing labels and more as different feedback architectures. Predictive delivery emphasizes control after plan deviation is detected; agile delivery emphasizes learning before uncertainty accumulates too far downstream. A balanced interpretation is important: predictive approaches can be appropriate where scope stability, regulatory control, and upfront coordination dominate; agile approaches are stronger where uncertainty, evolving stakeholder needs, and incremental value delivery dominate. The value of the SFDs is that they make these assumptions visible. They help practitioners discuss not only which approach to use, but what kind of feedback system the project actually needs.

References

Project Management Institute. (2026). PMBOK Guide. https://www.pmi.org/standards/pmbok

Project Management Institute. (n.d.). Project Performance Domains: Development Approach & Life Cycle. https://www.pmi.org/-/media/pmi/documents/public/pdf/pmbok-standards/pmbok-project-performance-domains.pdf

Project Management Institute. (2017). Agile Practice Guide. https://www.pmi.org/standards/agile

Sterman, J. D. (2001). System Dynamics Modeling: Tools for Learning in a Complex World. California Management Review, 43(4), 8-25. https://faculty.sites.iastate.edu/tesfatsi/archive/tesfatsi/SystemDynamics.JohnSterman2001.pdf